7.2. Angular Spreading and External shocks
Comparison of Eqs. 29 and 30 (using RE RE) reveals that Tang Tradial. As long as the shell's angular width is larger than -1, any temporal structure that could have arisen due to irregularities in the properties of the shell or in the material that it encounters will be spread on a time given by Tang. This means that Tang is the minimal time scale for the observed temporal variability: T Tang.
Comparison with the intrinsic time scales yields two cases:
In Type-I models, the duration of the burst is determined by the emission radius and the Lorentz factor. It is independent of . This type of models include the standard "external shock model" [27, 18, 233] in which the relativistic shell is decelerated by the ISM, the relativistic magnetic wind model  in which a magnetic Poynting flux runs into the ISM, or the scattering of star light by a relativistic shell [234, 235].
In Type-II models, the duration of the burst is determined by the thickness of the relativistic shell, (that is by the duration that the source is active and produces the relativistic wind). The angular spreading time (which depends on the the radius of emission) is shorter and therefore irrelevant. These models include the "internal shock model" [28, 29, 30], in which different parts of the shell are moving with different Lorentz factor and therefore collide with one another. A magnetic dominated version is given by Thompson .
The majority of GRBs have a complex temporal structure (e.g. section 2.2) with T / T of order 100. Consider a Type-I model. Angular spreading means that at any given moment the observer sees a whole region of angular width E-1. Any variability in the emission due to different conditions in different radii on a time scale smaller than Tang is erased unless the angular size of the emitting region is smaller than E-1. Thus, such a source can produce only a smooth single humped burst with = 1 and no temporal structure on a time-scale T < T. Put in other words a shell, of a Type-I model, and with an angular width larger than E-1 cannot produce a variable burst with >> 1. This is the angular spreading problem.
On the other hand a Type-II model contains a thick shell > RE / E2 and it can produce a variable burst. The variability time scale, is again limited T > Tang but now it can be shorter than the total duration T. The duration of the burst reflects the time that the "inner engine" operates. The variability reflects the radial inhomogeneity of the shell which was produced by the source (or the cooling time if it is longer than / c). The observed temporal variability provides an upper limit to the scale of the radial inhomogeneities in the shell and to the scale in which the "inner engine" varies. This is a remarkable conclusion in view of the fact that the fireball hides the "inner engine".
Can an external shock give rise to a Type-II behavior? This would have been possible if we could set the parameters of the external shock model to satisfy RE 2E c T. As discussed in 8.7.1 the deceleration radius for a thin shell with an initial Lorentz factor is given by
and the observed duration is l -8/3 / c. The deceleration is gradual and the Lorentz factor of the emitting region E is similar to the original Lorentz factor of the shell . It seems that with an arbitrary large Lorentz factor we can get a small enough deceleration radius RE. However, Eq. 34 is valid only for thin shells satisfying > l-8/3 . As increases above a critical value c = (l / )3/8 the shell can no longer be considered thin. In this situation the reverse shock penetrating the shell becomes ultra-relativistic and the shocked matter moves with Lorentz factor E = c < which is independent of the initial Lorentz factor of the shell . The deceleration radius is now given by RE = 1/4 l3/4, and it is independent of the initial Lorentz factor of the shell. The behavior of the deceleration radius RE and observed duration as function of the shell Lorentz factor is given in Fig. 15 for a shell of thickness = 3 × 1012cm. This emission radius RE is always larger than / E2 - thus an external shock cannot be of type II.
Figure 15. The deceleration radius Re and the Lorentz factor of the shocked shell e as functions of the initial Lorentz factor , for a shell of fixed width = 3 × 1012 cm. For low values of , the shocked material moves with Lorentz factor e ~ . However as increases the reverse shock becomes relativistic reducing significantly the Lorentz factor e < . This phenomena prevents the "external shock model" from being Type-II.